Giving exams to large numbers of students…

Giving a midterm later today has me thinking about the future of higher education. On the one hand, when remembering my very best classes at Amherst College, I recall a well-delivered lecture as a real gem, to be savored over time, potentially life changing. On the other, the logistics of handling learning assessment metrics (such as exams) for large classes, in the era before the Internet, created very real limitations, even at an elite liberal arts school like Amherst.

My worries about for-profit distance education models principally center around the genericness of the industrial scale lecture. But there may well be new frontiers in learning assessment leveraged by technology that might be usefully appropriated here in the non-profit university world. Ideally (and this is dream-ware I realize), one would want a way to use strong AI (perhaps in the context of natural language processing) to assist human professors in the grading of exams that include essays, equations and other non-multiple-choice instruments–Exams that we might have given comfortably in a seminar class of 15, but that are currently impossible in a class of 100 students.

My favorite examination at Amherst, was in quantum chemistry–it was one week long and open book. All of the questions involved sophisticated mathematics at the very limit of my education. In fact, much of the required mathematics was taught to us by the course professor so that we could understand the chemistry! Our professor was available by phone during that week pretty much 24/7. It was an incredibly exhilarating experience to pass that exam and it was about as far removed from multiple choice as one could possibly get.

My goal would be for technology to drive us towards learning experiences and assessments like the above. If we could do something like that, then we might really revolutionize higher education.